Zhiyong Zhang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Zhiyong Zhang has a total of 48 co-authors. Here are the top co-authors by number of shared publications:
Changzhen Qiu
10 shared publications- Changzhen Qiu
School of Electronics and Communication Engineering, Sun Yat-sen University, Shenzhen 518107, China
10 shared publications Shaolong Chen
School of Sino-German Intelligent Manufacturing, Shenzhen City Polytechnic, Shenzhen, China
5 shared publicationsCongjing Yu
4 shared publications- Shaolong Chen
School of Sino-German Intelligent Manufacturing, Shenzhen City Polytechnic, Shenzhen, China
4 shared publications Jing Ye
3 shared publications
Latest Publications
- 5 May 2026
- Published by Springer Nature in Signal, Image and Video Processing
- Vol. 20 (6) , 319
Multi-level Aggregation Attention RGB–Event Fusion Network for Moving Object Detection
- 5 May 2026
- Published by Springer Nature in Signal, Image and Video Processing
- Vol. 20 (6) , 319
Multi-level Aggregation Attention RGB–Event Fusion Network for Moving Object Detection
- Open Access
A Scene Detection Complexity Metric for Infrared Small Target Detection
Infrared small target detection is widely used in aerospace surveillance, maritime search and rescue, and military reconnaissance. However, the performance of detection algorithms is highly dependent on scene characteristics, and methods that perform well in simple backgrounds may degrade substantially in complex environments. Existing indicators, such as information entropy, average gradient, and peak signal-to-noise ratio, can reflect detection difficulty from individual perspectives, but they do not provide a unified measure that jointly considers target saliency, background complexity, and target–background coupling. To address this issue, this study proposes a scene detection complexity (SDC) metric for quantifying the difficulty of infrared small target detection. Six basic indicators are selected from three dimensions, namely target saliency, background complexity, and target–background coupling: statistical variance, target–background contrast, signal-to-clutter ratio, information entropy, structural similarity, and target size. After Min–Max normalization, objective weights are determined by combining the entropy weight method and principal component analysis, and the weighted indicators are fused into an SDC value in the range of . Experiments on 100 test images selected from IRST640, MSISTD, SIRST-V2, and an infrared small-aircraft sequence dataset show that the proposed SDC achieves a Pearson linear correlation coefficient of 0.956 with subjective difficulty ratings and with image-level detection scores obtained from seven representative algorithms. The results further indicate that traditional methods are more sensitive to increasing scene complexity, whereas deep-learning-based methods are comparatively more robust in complex backgrounds. The proposed SDC provides a unified and objective tool for performance evaluation, algorithm selection, and pre-assessment of scene difficulty in infrared small target detection.Open AccessA Scene Detection Complexity Metric for Infrared Small Target Detection
Infrared small target detection is widely used in aerospace surveillance, maritime search and rescue, and military reconnaissance. However, the performance of detection algorithms is highly dependent on scene characteristics, and methods that perform well in simple backgrounds may degrade substantially in complex environments. Existing indicators, such as information entropy, average gradient, and peak signal-to-noise ratio, can reflect detection difficulty from individual perspectives, but they do not provide a unified measure that jointly considers target saliency, background complexity, and target–background coupling. To address this issue, this study proposes a scene detection complexity (SDC) metric for quantifying the difficulty of infrared small target detection. Six basic indicators are selected from three dimensions, namely target saliency, background complexity, and target–background coupling: statistical variance, target–background contrast, signal-to-clutter ratio, information entropy, structural similarity, and target size. After Min–Max normalization, objective weights are determined by combining the entropy weight method and principal component analysis, and the weighted indicators are fused into an SDC value in the range of . Experiments on 100 test images selected from IRST640, MSISTD, SIRST-V2, and an infrared small-aircraft sequence dataset show that the proposed SDC achieves a Pearson linear correlation coefficient of 0.956 with subjective difficulty ratings and with image-level detection scores obtained from seven representative algorithms. The results further indicate that traditional methods are more sensitive to increasing scene complexity, whereas deep-learning-based methods are comparatively more robust in complex backgrounds. The proposed SDC provides a unified and objective tool for performance evaluation, algorithm selection, and pre-assessment of scene difficulty in infrared small target detection. - 4 February 2026
- Published by Elsevier in Biomedical Signal Processing and Control
- Vol. 118
AMF-MedIT: An efficient align-modulation-fusion framework for medical image–tabular data
- 4 February 2026
- Published by Elsevier in Biomedical Signal Processing and Control
- Vol. 118
AMF-MedIT: An efficient align-modulation-fusion framework for medical image–tabular data
- 13 January 2026
- Published by Springer Nature
- p. 292-305
AMC-SORT: Apply More Cues to Multi-Object Tracking
- 13 January 2026
- Published by Springer Nature
- p. 292-305
AMC-SORT: Apply More Cues to Multi-Object Tracking
- Open Access
- 27 October 2025
- Published by Association for Computing Machinery (ACM)
- p. 4629-4638
DSP: Dense-Sparse Parallel Networks for Self-supervised 3D Multi-person Pose Estimation from Multiple Views
- 27 October 2025
- Published by Association for Computing Machinery (ACM)
- p. 4629-4638
Open AccessDSP: Dense-Sparse Parallel Networks for Self-supervised 3D Multi-person Pose Estimation from Multiple Views
- Open Access
- 19 August 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1) , 1-17
A fully automatic knee subregion segmentation network based on tissue segmentation and anatomical geometry
Aiming at the difficulty of knee MRI bone and cartilage subregion segmentation caused by numerous subregions and unclear subregion boundary, a fully automatic knee subregion segmentation network based on tissue segmentation and anatomical geometry is proposed. Specifically, first, we use a transformer-based multilevel region and edge aggregation network to achieve precise segmentation of bone and cartilage tissue edges in knee MRI. Then, we designed a fibula detection module, which determines the medial and lateral of the knee by detecting the position of the fibula. Afterwards, a subregion segmentation module based on boundary information was designed, which divides bone and cartilage tissues into subregions by detecting the boundaries. In addition, in order to provide data support for the proposed model, fibula classification dataset and knee MRI bone and cartilage subregion dataset were established respectively. Testing on the fibula classification dataset we established, the proposed method achieved a detection accuracy of 1.000 in detecting the medial and lateral of the knee. On the knee MRI bone and cartilage subregion dataset we established, the proposed method attained an average dice score of 0.953 for bone subregions and 0.831 for cartilage subregions, which verifies the correctness of the proposed method.
- 19 August 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1) , 1-17
Open AccessA fully automatic knee subregion segmentation network based on tissue segmentation and anatomical geometry
Aiming at the difficulty of knee MRI bone and cartilage subregion segmentation caused by numerous subregions and unclear subregion boundary, a fully automatic knee subregion segmentation network based on tissue segmentation and anatomical geometry is proposed. Specifically, first, we use a transformer-based multilevel region and edge aggregation network to achieve precise segmentation of bone and cartilage tissue edges in knee MRI. Then, we designed a fibula detection module, which determines the medial and lateral of the knee by detecting the position of the fibula. Afterwards, a subregion segmentation module based on boundary information was designed, which divides bone and cartilage tissues into subregions by detecting the boundaries. In addition, in order to provide data support for the proposed model, fibula classification dataset and knee MRI bone and cartilage subregion dataset were established respectively. Testing on the fibula classification dataset we established, the proposed method achieved a detection accuracy of 1.000 in detecting the medial and lateral of the knee. On the knee MRI bone and cartilage subregion dataset we established, the proposed method attained an average dice score of 0.953 for bone subregions and 0.831 for cartilage subregions, which verifies the correctness of the proposed method.
- 8 August 2025
- Published by Elsevier in Pattern Recognition
- Vol. 171
STGFormer: Spatio-temporal GraphFormer for 3D human pose estimation in video
- 8 August 2025
- Published by Elsevier in Pattern Recognition
- Vol. 171
STGFormer: Spatio-temporal GraphFormer for 3D human pose estimation in video